Based on simulations with a commercial flowsheet simulator, optimized flowsheets are generated. Since the simulator does not provide explicit equations or derivatives, a memetic algorithm is used. This algorithm combines an evolutionary algorithm with a local derivative-free search. Discrete design variables and continuous parameters are optimized. Based on a case study, the behavior of the optimizer is demonstrated. Initial solutions generated by experts can be improved significantly by this approach but the computational time for complex flowsheets is still relatively high.